Hello everyone,
I am currently implementing state-space models in a real time control system for a project involving an autonomous mobile robot. I have covered the theoretical topics associated with state-space models (deriving the A, B, C, D matrices, stability analysis etc.) but have come across some practical challenges within the implementation process.
Specifically, I am interested in the best way to account for model inaccuracies and possible external disturbances in real-time. Should I tune an observer (like a Luenberger observer) for each environment I expect to operate in, or is a Kalman filter a better fit despite computational cost?
I am also interested in how people dealt with discretization when executing continuous-time state-space models on embedded systems. If you have any recommendations for methods, techniques or best practices for compromising accuracy and processing speed, I would appreciate it!
Any thoughts, suggestions or insight from anyone who has implemented these techniques in servicenow certification or robotics or an equivalent application would be welcome. I am trying to connect theory to real world behavior better, and I think your experiences could be valuable.
Thanks in advance!
williamclark
